568 Chapter 13:Quality Control
To determine the variance, we again use Equation 13.6.2:
Var(Wt)=
σ^2
n
{
α^2 +[α(1−α)]^2 +[α(1−α)^2 ]^2 +···+[α(1−α)t−^1 ]^2
}
=
σ^2
n
α^2 [ 1 +β+β^2 +···+βt−^1 ] whereβ=(1−α)^2
=
σ^2 α^2 [ 1 −(1−α)^2 t]
n[ 1 −(1−α)^2 ]
=
σ^2 α[ 1 −(1−α)^2 t]
n(2−α)
Hence, whentis large we see that, provided that the process has remained in control
throughout,
E[Wt]=μ
Var(Wt)≈
σ^2 α
n(2−α)
since (1−α)^2 t≈ 0
Thus, the upper and lower control limits forWtare given by
UCL=μ+ 3 σ
√
α
n(2−α)
LCL=μ− 3 σ
√
α
n(2−α)
Note that the preceding control limits are the same as those in a moving-average control
chart with spank(after the initialkvalues) when
3 σ
√
nk
= 3 σ
√
α
n(2−α)
or, equivalently, when
k=
2 −α
α
or α=
2
k+ 1
EXAMPLE 13.6b A repair shop will send a worker to a caller’s home to repair electronic
equipment. Upon receiving a request, it dispatches a worker who is instructed to call in
when the job is completed. Historical data indicate that the time from when the server
is dispatched until he or she calls is a normal random variable with mean 62 minutes
and standard deviation 24 minutes. To keep aware of any changes in this distribution,